MIT's Project NANDA (2025) found that 95 percent of enterprise generative AI initiatives produced no measurable effect on profitability. Harvard Business Review research confirms that cognitive exhaustion from AI agent oversight is real and significant. The specific mechanisms driving burnout remain partially unclear—whether the issue stems primarily from token costs, workflow disruption, or the psychological toll of oversight work. The extent to which individual firms are tracking or responding to these burnout signals is not yet documented.
Attorneys should monitor this trend closely. If burnout translates into attrition, firms may face talent retention litigation, wage-and-hour claims tied to unpaid oversight work, or disability claims related to cognitive exhaustion. The disconnect between AI adoption pressure and actual business returns creates liability exposure for companies making aggressive AI implementation commitments to investors or clients. Employment counsel should begin documenting how clients are structuring AI workflows and whether adequate rest and creative autonomy are being preserved in development cycles.